Predicting ammonia and carbon dioxide emissions from carbon and nitrogen biodegradability during animal waste composting

ARTICLE IN PRESS

Atmospheric Environment 39 (2005) 6833–6842 www.elsevier.com/locate/atmosenv

Predicting ammonia and carbon dioxide emissions from carbon and nitrogen biodegradability during animal waste composting Jean-Marie Paillata,, Paul Robinb, Me´lynda Hassounab, Philippe Letermec a

Cirad, UMR SAS, 65 rue de St Brieuc, cs84215, 35042 Rennes cedex 01, France Inra, UMR SAS, 65 rue de St Brieuc, cs84215, 35042 Rennes cedex 01, France c Agrocampus, UMR SAS, 65 rue de St Brieuc, cs84215, 35042 Rennes cedex 01, France b

Received 8 April 2005; received in revised form 13 July 2005; accepted 26 July 2005

Abstract During composting of livestock manure, transformations of organic matter result in gaseous emissions, which can harm the environment. Two experiments were done in enclosures to measure the fluxes of NH3, N2O, CO2, CH4 and H2O emitted by 8 heaps of compost representing the range of biodegradability of nitrogen and carbon in the livestock manure. The heaps were monitored for the first 2 months, corresponding to the thermophilic phase during which nearly all-mass losses occur. Four parameters describe the NH3 emission kinetics and the main influential factors were noted: (1) the response time to reach maximum intensity is affected mainly by the initial micro-flora; (2) the amplitude depends mainly on C biodegradability and also on micro-flora; (3) the emission duration depends mainly on N biodegradability; and (4) the cumulative emission, which varied from 16.5 to 48.9% of the nitrogen initially present in the heap, depends both on C and N biodegradability. A predictive model for NH3 and CO2 emissions for the thermophilic phase of the composting of livestock manure is proposed. The variability in cumulative emissions of CO2 and of NH3 is well explained by the contents of soluble elements and hemicellulose in the dry matter (Van Soest fractioning), and soluble nitrogen (12 h extraction at 4 1C in water). In our conditions of favourable aeration and humidity, N2O and CH4 emissions were low. The role of the biodegradable carbon in reducing NH3 emission is highlighted. r 2005 Elsevier Ltd. All rights reserved. Keywords: Ammonia; Carbon dioxide; Greenhouse gases; Compost; Animal manure

1. Introduction Composting is a useful technique to facilitate the management of livestock manure, which is one of the major problems in regions of intensive livestock Corresponding author. Tel.: +33 0 223485431;

fax: +33 0 223485430. E-mail address: [email protected] (J.-M. Paillat).

production (Mustin, 1987). However it leads to large gaseous emissions, notably of NH3, inducing losses of elements and harmful effects on the environment. NH3 is responsible for acidification of rain and of the environment and for the formation of aerosols (Apsimon et al., 1987; Fangmeier et al., 1994), N2O and CH4, which are powerful greenhouse effect gases (Kroeze, 1994; Houghton et al., 2001). Few studies have attempted to quantify the relationship between gaseous emissions during composting and

1352-2310/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2005.07.045

ARTICLE IN PRESS J.-M. Paillat et al. / Atmospheric Environment 39 (2005) 6833–6842

6834

the biochemical forms of nitrogen and carbon present in the composted material (Andersson, 1996; Pare´ et al., 1998). Kirchmann and Witter (1989) fitted a parallel first-order model, based on the decomposition of a rapidly and a slowly decomposable nitrogenous fraction, on NH3 volatilisation data and parameterised it for C:N ratio of 18, 24 and 36. Eklind and Kirchmann (2000) parameterised an organic matter decomposition model for organic household waste and predict organic C losses by distinguishing the fractions according to their biodegradability. Liang et al. (2004) proposed a model based on microbial growth and heat production. Ekinci et al. (2000) predict NH3 emission from the C:N ratio and the pH. To assess the effects of C and N availability and microbial flora on the potential emissions, we carried out two experiments with different composting situations for manure in which the biodegradable C and N contents in the initial material were varied, in favourable and similar conditions of water content and initial porosity. The objectives of this article are (1) to propose a predictive model for the NH3 and CO2 emissions during the thermophilic phase of composting, based on the biodegradable C and N contents in the initial material; (2) to precise the role of the flora on the dynamics of the emissions; and (3) to provide data for NH3, N2O and CH4 emissions from composting process.

2. Materials and methods 2.1. Materials and experimental design We used organic matter fractionating (Van Soest, 1963) to quantify C biodegradability. Robin (1997) combined the different Van Soest fractions to predict the biological stability of substrates; Morvan et al. (2001) explained carbon and nitrogen mineralisation of organic

products in the soil in terms of these fractions. Eklind and Kirchmann (2000) found that the lignin-like compartment is the best predictor of residual organic C after the whole process of composting (590 days). Since, CO2 and NH3 emissions occur mainly in the first period of the aerobic composting process (Beck–Friis et al., 2001), we assumed that soluble and hemicelluloselike fractions (SH-VS) should well characterise the initial biodegradable C. To estimate the biodegradable N, as done for silage analysis (Dulphy and Demarquilly, 1981), we considered the proportion of soluble nitrogen (SN) in the total nitrogen (TN); this fraction is composed of the ammonium and the organic N in the solution, in particular the non-hydrolysed urea; no nitrate or nitrite were found in the initial materials (o0.01 g kg
1 ww). In experiment 1, four mixtures, A, B, C and D (Tables 1 and 2) composed of pig manure, pig slurry, urea, wheat straw and water, were made in order to obtain a range of biodegradable N contents: SN:TN varied from 44% to 87% (Fig. 1). In experiment 2, four mixtures E, F, G and H (Tables 1 and 2) composed of wheat straw, sawdust, pig slurry, sugar beet molasses, urea and water were made to obtain varying biodegradable C contents. The SH-VS varied from 51% to 73% of DM (Fig. 1). The eight heaps had the same shape (half a swath), volume (1.3770.04 m3) except for H (1.18 m3) and mass (409712 kg). Water content (70.271.4%), density (305720 kg m
3) and free air space inside the compost heap (0.7370.02) were fairly similar. Each heap was made in one enclosure of 8 m3, thermally insulated (Fig. 2), placed in a thermally insulated building containing four enclosures. The temperature inside the building was kept below 13 1C and the temperature inside the enclosures was above 25 1C; heat was provided by the compost and with an electrical heater when the heap temperature fell. By adjusting the openings, the air flow rate inside each

Table 1 Composition (% on wet weight basis) of the composting mixtures Heap identification

A

B

C

D

E

F

G

Hb

Wheat straw Sawdust Sugarcane molasses Pig manure Pig slurrya Urea Water

18.5 — — 45.0 36.5a — —

28.1 — — 15.3 — 0.5 56.1

10.0 — — 80.1 — 0.4 9.5

26.0 — — — 72.8a 1.2 —

21.0 — — — 78.9b 0.1 —

12.7 10.6 — — 52.9b 0.6 23.2

15.3 — 7.5 — 71.9b — 5.3

19.4 2.2 10.9 — 15.6b 0.8 51.1

a

The pig slurry had a very different chemical composition in the two experiments (dry matter contents of 0.9 vs 10% of the gross weight for experiment 1: A,B,C and D heaps, and experiment 2: E,F,G and H heaps, respectively). b Because of the large amount of water, compared with the coarse material, mixed in initially when making the heaps, this mixture lost a lot of liquid containing part of the incorporated soluble elements (sugars, mineral N); we therefore do not show here the characteristics of the different products, in so far as it is the composition of the mixtures (Table 2), which is important in determining the gaseous emissions.

ARTICLE IN PRESS J.-M. Paillat et al. / Atmospheric Environment 39 (2005) 6833–6842

6835

Table 2 Chemical and biochemical composition of one homogeneous sample representative of each composting heap (error estimation ¼ 5%) Heap identification

A

B

C

D

E

F

G

H

a


1

306 289 288 324 308 290 297 282 DM content (g kg ww) TN (g kg
1 ww)b 4.2 4.8 6.9 7.1 7.2 6.7 6.6 6.5 1.9 2.6 3.7 6.2 4.8 5.0 4.2 4.9 SN (g kg
1 ww)c TC (g kg
1 ww)d 123 119 121 136 141 141 131 127 pHe 8.8 (0.3) 9.1 (0.3) 8.8 (0.4) 9.0 (0.2) 8.6 (0.1) 9.2 (0.3) 7.5 (0.8) 8.4 (0.4) C:N 29.0 24.7 17.6 19.0 19.5 21.2 19.6 19.5 Van Soest fractions (% DM)f Soluble (100–NDF) 34.0 28.7 32.5 24.9 37.3 27.6 53.2 41.3 Hemicellulose-like (NDF–ADF) 26.9 27.9 26.8 27.6 26.6 23.2 19.7 23.0 Cellulose-like (ADF–ADL) 32.9 36.8 34.2 41.6 31.0 38.7 23.7 29.6 Lignin-like (ADL) 6.3 6.6 6.5 5.9 5.1 10.4 3.4 6.1 a

Dry matter. Total nitrogen: Kjeldahl method. c Soluble nitrogen: Kjeldahl method on the extracted juice from maceration of 25 g of fresh material in 200 g of deionised water at 4 1C for 12 h. d Total carbon: Dumas method. e values in parenthesis are standard errors (n ¼ 9). f NDF: Neutral detergent fiber; ADF: Acid detergent fiber; ADL: Acid detergent lignin. b

SH-VS:DM

0.75

G

0.65

E A

H

C B

0.55

D

F 0.45 0.4

0.5

0.6

0.7

0.8

0.9

SN:TN Fig. 1. Classification of the different mixtures studied in relation to biodegradable nitrogen (SN:TN) and to biodegradable carbon (SH-VS:MS); SN: soluble nitrogen, TN: total nitrogen, SH-VS: soluble+hemicellulose-like fractions from Van Soest analysis (1963), DM: dry matter.

enclosure was regulated and varied between 0.3 and 0.8 m s
1 in order to achieve a regular static ventilation due to convection (chimney effect). The air inside and outside the four enclosures was automatically pumped and analysed every 2 min during 15 min location
1; the duration of a measure cycle was 90 min (six sampling points). A data logger (SA120 from AOIP SAS, Ris-Orangis F-91130) was used to switch the pumps. For this 15 min period, we calculated for each gas an average concentration based on four stable consecutive measurements. Hence, sampled air could not be contaminated by surrounding air. The concentration measurements were monitored using three devices: the 3426 analyser from Bru¨el & Kjaer for

NH3, N2O and H2O, the 1312 analyser from Innova for NH3, CO2, CH4 and H2O (Bru¨el & Kjaer, Innova, Skordsborgvej 307, Naerum DK-2850), and an infrared sensor for CO2 (Gascard II Edinburgh Sensors, Kirkton Campus, Livingston, EH 54 7DQ, UK). The concentration measurements were automatically recorded by a computer (RS232 interface). In order to validate the continuous analysers measurements for CO2 and NH3 concentrations, manual measurements were made regularly for the six sampling points with colorimetric tubes (Dra¨ger Safety France SAS, Strasbourg F-67025 cedex). Dra¨ger tubes were also used to check that the NOx concentrations remained negligible. Water vapour concentrations and air densities were calculated from wet and dry temperatures provided by psychrometers fitted with type K thermocouples. Air speed was measured continuously for each enclosure through one of the upper-calibrated holes using a hotwire anemometer (8450 from TSI Incorporated, Shoreview, MN 55126 USA). Using a rotating vane anemometer (LCA 6000 from Airflow Developments LTD, Cressex Business Park, High Wycombe, Buckinghamshire, HP12 3QP, UK), manual measurements were made in each opened hole, twice a day at the beginning of the experiment and every day after the temperature rise in the compost, to correct the possible errors of TSI measurements due to air speed heterogeneity between holes. The mass balance of water was used to check the absence of bias in the emission measurements. Temperatures inside the heaps were monitored by type K thermocouples located on one side, at the bottom, in the middle and on top (two thermocouples in each position). Temperatures inside the heaps, dry and wet air temperatures inside and

ARTICLE IN PRESS J.-M. Paillat et al. / Atmospheric Environment 39 (2005) 6833–6842

6836

Gas concentration : N2O, CO2, NH3, CH4, H2O

Computer

Air outlet

Datalogger

Air speed

Inside air

Air temperature, humidity

Outside air

Air temperature, humidity

Datalogger

Air inlet

Compost temperature

Compost

Fig. 2. Schematic representation of the device used for measuring gaseous emissions.

outside the enclosure and air speeds were recorded every 2 min and averaged every 30 min by two data-loggers (SA70 from AOIP, SAS, Ris-Orangis F-91130, and 21X from Campbell Scientific, Courtaboeuf, F-91967 cedex). 2.2. Compost sampling and analytical methods The different mixtures (Table 1) were mixed several times by a rotary cultivator at day 0 and removed from the enclosure at 61 and 92 days for experiments 1 and 2, respectively. At each date, the heap was weighed and sampled following the protocol of the French Energy and Environmental Agency (Ademe, BP 90406, Angers F-49004 cedex 01): 25 samples of 300–500 g were collected from each heap and mixed in an aggregated sample which was then manually homogenised and divided in two parts several times until yielding a final sample of 1 kg. The final samples (one for each treatment) were deep frozen (
18 1C) immediately. Each frozen sample was chopped and ground with a mixer (Robotcoupe blixer 5+ from Ecotel, Vezin le Coquet F35132), then divided into four sub-samples corresponding to the intended analyses and immediately refrozen to avoid gaseous losses. One of the four sub-samples was dried at 40 1C (200 g), two were conserved as fresh

material (2  10 g) and one was macerated with deionised water (25:200) in a closed vessel at 4 1C for 12 h. The remaining material was kept at
18 1C. This sample treatment led to a good representation of the compost heap: following this protocol, a preliminary experiment showed that the coefficients of variation of analytical data were less than 5%. Dry matter content was calculated by oven drying at 40 1C until a constant weight was reached. This dried material was ground to 1 mm for the Van Soest analysis (1963). This same material and each Van Soest fraction were finely ground to assess the total C content (Dumas oxidation method ISO 10694-1995 with Flash 1112 analyser from Thermo Electron corp., Courtaboeuf F91962 cedex). The fresh material was analysed for total N by Bu¨chi distillation after mineralisation by the Kjeldahl method (ISO 5663-1994) and for ammoniacal N by Bu¨chi distillation (fresh material in deionised water at a ratio of 5:150; NF T90015-1975). The material resulting from maceration was centrifuged and filtered (0.45 mm); SN was estimated by total N measured on this extracted juice by Bu¨chi distillation after mineralisation by the Kjeldahl method (Dulphy and Demarquilly, 1981); nitric nitrogen (nitrate+nitrite) was also measured on this extracted juice by RFA300 auto-analyser

ARTICLE IN PRESS J.-M. Paillat et al. / Atmospheric Environment 39 (2005) 6833–6842

from Alpkem corp. (Clackamas, OR 97015, USA). Just after the heap making, the pH was measured at 9 points (at a depth of 5 cm in the fresh material) within the whole heap surface with a Xerolyt probe from Ingold (Mettler-Toledo AG, lm Langacher, CH-8606 Greifensee) connected to a 320 pH-meter from WTW (Champagne´ au Mont d’Or F-69410). 2.3. Calculation and device accuracy The ammonia emissions were calculated at each time step using the following expression: E NH3 -N ¼ ðnh Ah S  3600Þ  rTdi;Twi   MN 1   ðC i
C o Þ  , N init rnorm  M NH3

ð1Þ

where E NH3 -N was the ammonia emission of one enclosure (g NH3-N h
1 kg
1 initial N), nh was the number of opened holes, Ah was the area of each hole (m2), S was the mean air speed of the output air through the holes (m s
1), rTdi,Twi was the outlet air density (kg dry air m
3 humid air) calculated from the dry and wet-bulb temperatures, Ci and Co were the observed concentrations inside and outside the enclosure (g NH3 m
3 at normal temperature and air pressure), MN and M NH3 were the molar masses (g N mol
1 and g NH3 mol
1, respectively), rnorm was the density of normal air (kg dry air m
3 normal air), Ninit was the initial nitrogen mass of the heap (kg N). For water vapour we calculated the water concentration in kg water kg
1 dry air directly from the dry and wet-bulb temperatures. For the other gases (CO2, CH4, N2O) we adapted the molar masses and the initial nitrogen mass in Eq. (1). We estimated the uncertainty of the emissions by comparing (1) the concentrations measured by colorimetric tubes and the gas analyser; (2) the concentrations measured inside and outside the enclosures when they were empty, and (3) the air speed measured by the hotwire anemometer and the rotating vane anemometer. The emission of CO2 was also compared to the sensible and latent energy emitted by the enclosure, using the approximate ratio of 0.163 L CO2 h
1 W
1 used in

6837

livestock production (CIGR, 1984). The uncertainty of the total emission was around 10% of the value. It was higher for gases with low emissions such as N2O and CH4 when there is a small concentration gradient between inside and outside air. We checked that this uncertainty estimate was consistent with the cumulative emission and mass loss after 61 and 92 days for experiments 1 and 2, respectively. In another experiment (unpublished), repeating the same composting in two enclosures over 14 days, we observed around 10% difference between the emissions and heat production of the two heaps. However, we only interpreted the differences of emissions between treatments when they were above 20%. In order to link NH3 and CO2 emissions to initial C and N biodegradability and to keep within the thermophilic phase, we calculated the balance of gaseous emissions at 28 and 56 days (Table 3), although the experiments lasted more than 56 days. Indeed, from 28 days after making the heaps, some heaps cooled markedly: the temperature fell below 50 1C and after 56 days all the heaps cooled below 50 1C (Fig. 3). We compared C and N left in the end product to the amount of emitted C and N at 61 and 92 days, respectively, for experiments 1 and 2, respectively. The difference in C recovery ranged from
1 to 11% of initial C content, except for A and C heaps with large amounts of pig manure incorporated (16 and 21%, respectively). Other volatile compounds could have been emitted. The difference in N recovery ranged from
6% to 17% of initial N. A small amount of N2 could have been emitted, particularly when the temperature fell. All data processing was done using Microsoft Excels.

3. Results 3.1. Kinetics of gaseous emissions and temperature The CO2-C, CH4-C, NH3-N and N2O-N emission kinetics can be compared for the eight heaps, with four parameters (Fig. 3): the cumulative emission, the time to

Table 3 Balance of gaseous emissions after 28 and 56 days of composting (error estimation ¼ 10%) Identification of the heap

Time (d)

A

B

C

D

E

F

G

H

NH3-N (g kg
1 TN) NH3-N (g kg
1 TN) CO2-C (g kg
1 TC) CO2-C (g kg
1 TC) N2O-N (g kg
1 TN) N2O-N (g kg
1 TN) CH4-C (g kg
1 TC) CH4-C (g kg
1 TC)

28 56 28 56 28 56 28 56

159 165 364 423 9.3 11.3 — —

270 275 365 448 8.6 10.6 — —

385 387 354 407 6.9 10.2 — —

478 489 285 388 5.1 7.1 — —

357 362 342 407 7.3 8.1 0.4 0.6

469 479 276 364 6.8 7.7 0.4 0.5

243 248 379 426 9.3 12.0 0.1 0.2

233 249 363 425 8.8 10.3 0.2 0.3

TN: total nitrogen; TC: total carbon.

ARTICLE IN PRESS J.-M. Paillat et al. / Atmospheric Environment 39 (2005) 6833–6842

6838

80

8 7

70

E

A

60

6 Time to reach the maximum emission

5

50

Amplitude of the emission peak

4

40 30

3 Cumulative emission

2

20 10

1

0

0 8

80

CO2-C emission (g kg-1 TC); NH3-N emission (g kg-1 TN)

7

B

70

F

6

60

5

50

4

40

3

30

2

20

1

10

0

0

8

80

NH3-N peak at 12,9 g kg-1 TN

7

C

70

G

6

60

5

50

4

40

3

30

2

20

1

10

0

0

8

80

7

D

CH4-C emission (mg kg-1 TC); N2O-N emission (mg kg-1 TN); temperature (˚C)

Duration of the emission peak

70

H

6

60

5

50

4

40

3

30

2

20

1

10 0

0 0

7

14

21

28 Time (d)

35

42

49

56

0

7

14

21

28

35

42

49

56

Time (d)

Fig. 3. Kinetics of gaseous emissions during the thermophilic phase of composting; for heaps A–H, CO2-C emission (g kg
1 TC), shown by a bold grey line, and NH3-N emission (g kg
1 TN), shown by a bold black line, refer to the left axis; for heaps E–H, CH4-C emission (mg kg
1 TC), shown by a crossed line, and for heaps A–H, N2O-N emission (mg kg
1 TN), shown by empty circles, refer to the right axis; the temperature (1C) in the centre of each heap, shown by dashes, refers to the right axis; TC is total carbon, TN is total nitrogen. The four parameters describing the NH3 emission kinetics are shown for heap E.

ARTICLE IN PRESS J.-M. Paillat et al. / Atmospheric Environment 39 (2005) 6833–6842

reach maximum emission, the amplitude and the duration of the emission peak. For a better understanding of the process involved, temperature is also shown on Fig. 3. In the presence of a less biodegradable substrate (F and D) the rise in temperature is slowed down (the maximum temperature was reached only after 12 days); conversely it is accelerated when the carbon is more biodegradable (8, 4.5, 3.8 for E, G and H, respectively), and when the micro-flora is present when the heap is built. Indeed, the amount of manure in the mixture directly influenced the rapidity of the temperature rise: 1.6, 1.8 and 2 days, respectively, for heaps C, A and B with 80, 45 and 15% of manure, respectively. The temperature plateau was around 70 1C for all the heaps. The steady decline in temperature thereafter corresponds to the fall in activity of thermophilic organisms and to the loss in heat by air heating and water evaporation. This decline was less marked for heaps D and F in which the activity was more prolonged because of the presence of nitrogen and the slower degradation of the less biodegradable carbon compounds. 3.2. Prediction of potential ammonia and carbon dioxide emissions Table 3 shows the cumulative emissions of CO2-C, CH4-C, NH3-N and N2O-N for the 28 and 56 days periods. For heaps E, F, G and H the cumulative emissions of CH4-C were very low and represent less than 0.2% of the carbon volatilised. The ratio between cumulative emissions of N2O-N and NH3-N at 56 days varied from 1.5:100 to 7.1:100. Heaps F and D which had the highest cumulative NH3-N emission had the lowest cumulative N2O-N emission. The biodegradability of both nitrogen estimated from SN content and carbon estimated from SH-VS content are necessary to predict NH3-N at 56 days. Probably because nitrogen is not limiting in the case of animal manure, only the C biodegradability estimated from SHVS content was necessary to predict CO2-C emissions at 28 and 56 days: cumulative NH3 -N emission at 56 days ¼ 16:38 SN
0:903 SH-VS þ 643:7 2

ðN ¼ 8; Po0:05; R ¼ 0:82Þ,

ð2Þ

cumulative CO2 -C emission at 28 days ¼ 0:683 SH-VS
58:92 ðN ¼ 8; Po0:01; R2 ¼ 0:84Þ,

ð3Þ

cumulative CO2 -C emission at 56 days ¼ 0:486 SH-VS þ 131:6 ðN ¼ 8; Po0:05; R2 ¼ 0:62Þ,

ð4Þ

6839

where NH3-N is expressed in g kg
1 TN, CO2-C in g kg
1 TC, SN and SH-VS in g kg
1 DM (TN, TC are total nitrogen and total carbon, respectively). The standard error (SE) for NH3-N emission at 56 days (Eq. (2)) is 58.6 g NH3-N kg
1 TN, or 18% of the mean emission. SE for CO2-C emissions at 28 d (Eq. (3)) and 56 days (Eq. (4)) are 23.0 and 29.3 g CO2-C kg
1 TC, i.e. 6.5% and 6.9% of the mean emission, respectively. Eq. (4) is less significant than Eq. (3). Indeed, cumulative emissions of CO2-C converge over time; the role of the biodegradability of the initial material becomes less important. According to Martins and Dewees (1992), the higher is the C:N ratio, the less is the volatilisation. Hence, Ekinci et al. (2000) use the C:N ratio and the pH to predict the ammonia emissions: cumulative NH3 -N emission ðg kg
1 TNÞ ¼
7:09 C : N þ 82:5 pH
203 ðN ¼ 27; R2 ¼ 0:9Þ.

ð5Þ

When applied to our data, this equation did not prove satisfactory, probably because the availability of the energy source (biodegradable C) to micro-organisms is not taken into account (Kirchmann, 1985): predicted NH3 -N emission ¼ 0:317 observed NH3 -N emission þ 257:5 ðN ¼ 8; Po0:05; R2 ¼ 0:63Þ.

ð6Þ

The prediction of NH3-N emissions is better with Eq. (2) which takes account of C and N biodegradability: predicted NH3 -N emission ¼ 0:818 observed NH3 -N emission þ 59:96 ðN ¼ 8; Po0:01; R2 ¼ 0:82Þ.

ð7Þ

4. Discussion 4.1. Factors influencing the dynamics of emissions The production of CO2 results mainly from the oxidation of easily degradable carbon compounds and consequently from the nature of the substrates, oxygen and moisture and the activity of micro-organisms (Andersson, 1996). The time to reach the emission peak is shortened with effective initial microbial flora brought in by the large amount of manure and slurry (0.5 days for heap A). The amplitude of the emission peak increases with carbon biodegradability: o1.5 or 42 g CO2-C kg
1 TC h
1 for D, F (SH-VS:MSo0.53), and for A, E, G, H (SH-VS:MS40.6), respectively. Heaps B and C are intermediate. After one month the intensity of the CO2 emission is similar in the different heaps

ARTICLE IN PRESS 6840

J.-M. Paillat et al. / Atmospheric Environment 39 (2005) 6833–6842

(0.2 g CO2-C kg
1 TC h
1), suggesting that the very biodegradable compounds have been used up. After two months, the cumulative emissions converge; consequently, the quantity of carbon remaining (56% to 64%) becomes almost constant, close to the observations of Pare´ et al. (1998) after 59 days (63%). The dynamics of NH3 emission were very different in the various heaps (Fig. 3). The response time to reach maximum intensity depends strongly on the quantities of manure and slurry, which determine initial microbial flora (Table 1). It is short with manure (0.5 to 1 day for A, B, C), intermediate with a lot of concentrated slurry (2 days for E, G) and long with little or slightly concentrated slurry (2.7 to 3.2 days for D, F, H). The abundance of flora and the C biodegradability also affect the amplitude of the emission peak. On the one hand, for comparable C biodegradability, heaps composed of 80% of manure+slurry (C, E) shows higher amplitudes (12.9 and 5.4 g NH3-N kg
1 TN h
1, respectively) than those with 15% of manure+slurry (B, H, with 4.2 and 1.7 g NH3-N kg
1 TN h
1). On the other hand, for a comparable quantity of slurry the amplitude for heap E (5.4 g NH3-N kg
1 TN h
1) is greater than for G (2.6 g NH3-N kg
1 TN h
1) due to the presence of molasses in G, which has enhanced the C biodegradability. We note also that for the same large quantity of biodegradable N, the amplitude for H (1.7 g NH3N kg
1 TN h
1) is lower than that for F (5.2 g NH3N kg
1 TN h
1) probably because of the higher C biodegradability in H. When the C biodegradability increases, the immobilisation of mineral N can be more intense and so effectively reduce NH3-N emission according to the assumption of Kirchmann and Witter (1989). The peak duration is closely linked to the available N (Fig. 1): 8 days for A, 10 for B and C, 12 for E and G, 16 for F and H, 25 for D. The cumulative NH3-N emission (Table 3) depends on the biodegradability of both C and N but little on the initial microbial flora. This fact explains why we can find a simple equation to predict the whole NH3 emission. The N2O is produced during nitrification and during incomplete denitrification (Mancinelli, 1992). As reported by Beck-Friis et al. (2001), the emission of N2O is low compared with that of NH3: in our experiment, the amplitude of the emission peak was 100 times less, except for A, G and H where this ratio was about 50:1. For all the heaps, a peak was observed in the early stage of composting: during the first day for A and C (made with a lot of decomposing farmyard manure) and between 1.6 and 2.4 days for the other heaps. Heaps A and C had the highest emission rates (57 and 53 mg N2O-N kg
1 TN h
1, respectively). As shown by He et al. (2000) the microbial flora can explain these highinitial emission rates. Heap D presented a low-emission peak (13 mg N2O-N kg
1 TN h
1) probably because urea which was the main source of nitrogen in this heap

was not totally hydrolysed when making the heap. For the other heaps, nitrification and perhaps simultaneous denitrification near the surface of the heap, where O2 is present and the temperature not too high, may have been responsible for this early production, as reported by Sommer and Mfller (2000). As soon as the temperature had risen, N2O production slowed down. Later, N2O was produced again when the temperature fell below 40 1C in heaps C, G and H, after approximately 45 days of composting. Methane emission was measured only during experiment 2. As Sommer (2001) found, the emission of CH4C was very low, in this experiment always at least 200 times less than that of CO2-C. The dynamics of emission are very similar to those observed for NH3 (Pel et al., 1997), these two gases being mainly emitted during the thermophilic phase (Hellmann et al., 1997; Fukumoto et al., 2003). Production of CH4 comes from microbial activity in anoxic areas of the heap, but most of the CH4 produced is oxidised by methanotrophic bacteria in the aerobic layers near the surface (Ja¨ckel et al., 2004; Wilshusen et al., 2004). 4.2. Ammonium fluxes within the heap Volatilisation of NH3 is the main route of nitrogen loss from livestock effluent compost (Kirchmann, 1985). When the heap is made, the ammoniacal nitrogen pool consists of ammonium ions present initially in the solution (slurry and manure), and these ions are from the ammonification of readily degradable nitrogen compounds depending on the micro-flora initially present and also from the hydrolysis of urea. Within the heap, the ammonium may remain in solution or be adsorbed onto the organic matter. From the ammonium pool, three consumptive fluxes are carbon dynamics dependent: nitrification, immobilisation by the microbial biomass and volatilisation (Kirchmann and Witter, 1989). Nitrification is conditioned by the presence of oxygen and available carbon; the nitrifying bacteria are sensitive to temperature (Hellmann et al., 1997). In the initial composting conditions, the high temperature and the CO2 produced prevent nitrification (Schlege, 1993). However in the external layers of the heap which dry out more rapidly and with lower temperatures, nitrites and nitrates can be produced but are probably rapidly assimilated (Mancinelli, 1992). Hence, the low rates of nitrification and denitrification lead to a low emission of N2O and prevent the accumulation of nitrates during the first two months of composting (a small amount of nitrate was measured in the end product: between 0.4 and 2.4% of TN). The immobilisation of the ammonium depends on the biodegradable C present and the activity of the microbial flora. The small ammonium pool present in

ARTICLE IN PRESS J.-M. Paillat et al. / Atmospheric Environment 39 (2005) 6833–6842

the end product (between 5 and 12%) confirms that immobilisation is probably the dominant flux. For similar C:N ratios, more biodegradable C (molasses) reduced NH3 emission, implying active immobilisation by the microbial biomass of the ammonium thus removed from volatilisation. Less biodegradable C (sawdust) increased NH3 emission, signifying less immobilisation by the microbial biomass. We therefore believe that the C:N ratio is not sufficient to explain the immobilisation of nitrogen during the composting of livestock manure. Hence the volatilisation flux of ammonia is related to the biodegradable nitrogen present and produced (Martins and Dewees, 1992) and to the two ‘‘sink’’ fluxes, namely nitrification (low in our case) and immobilisation, which depend not only on the C:N ratio (Kirchmann, 1985; Ekinci et al., 2000), but also on the C biodegradability (Andersson, 1996; Pare´ et al., 1998) and the adaptation of the flora to these substrates (Hellmann et al., 1997). NH3 emission into the atmosphere is moreover governed by the physico-chemical conditions, brought about by the transformations of the organic matter. With increasing temperature, the NH+ 4 :NH3 equilibrium shifts towards more NH3aq; the NH3aq:NH3g equilibrium shifts towards more NH3g and NH3 transfer to the atmosphere is increased (Peigne´ and Girardin, 2004). The pH also shifts the NH+ 4 :NH3 equilibrium towards more NH3aq. It is conditioned both by the dynamics of the organic matter degradation (production of NH+ 4 , bases and organic acids can lead to increasing or decreasing pH) and the emissions of NH3 and CO2. NH3 emission leaves H+ ions, which can be fixed by carbonates to produce CO2. The rapid acid degradation leads also to a pH increase (Beck-Friis et al., 2001). Therefore, the dynamics of pH in compost is very complex and variable as this parameter both influences and depends on the emissions. In our experiment, pH was high for all the heaps except for G and H where molasses resulted probably in high-organic acid production (Table 2). Then pH decreased markedly in heaps D and F with high-NH3 emission and increased slightly in heap G with high-CO2 emission. After a few days, variability in the measurements precluded calculation of statistical differences between heaps. For all heaps, pH remained between 7.2 and 7.9 after 10 days of composting. Emissions are also driven by the diffusion and convection of gas in the heap. These last parameters depend on the free porosity to the air (Shi et al., 1999; Ekinci et al., 2004; Liang et al., 2004; Chadwick, 2005), a factor not studied here.

5. Conclusion and prospects We have shown that, the C and N biodegradability can be quantified by measuring SN and SH-VS contents.

6841

These measurements enable the potential emission of NH3-N and CO2-C to be calculated for the thermophilic phase of composting of livestock manure, although the initial microbial flora influences the emission kinetics. The results show clearly the importance of immobilisation of nitrogen by the microbial biomass, which is the principal flux concurrent with ammonia volatilisation in the first weeks of composting livestock manure. This immobilisation is heavily dependent on the C biodegradability. The factors free air space and moisture content, which have been studied in other experiments, also strongly influence gaseous emissions. The four factors, biodegradable C and N, oxygen and moisture content, taken into account simultaneously in the same model, should be able to explain a significant part of the variations in gaseous emissions during the composting of livestock manure.

Acknowledgements We wish to thank the DEA students, the research collaborators and the laboratory technicians who participated in the work on composts. We also thank the reviewers for their very useful comments, which improved this paper. This work was made possible by funding from the ADEME within the framework of GIS Green Piggery and the provision by UMR SENAH of the experimental facilities.

References Andersson, 1996. Performance of bedding materials in affecting ammonia emissions from pig manure. Journal of Agricultural Engeneering Research 65 (1), 213–222. ApSimon, H.M., Kruse, M., Bell, J.N.B., 1987. Ammonia emissions and their role in acid deposition. Atmospheric Environment 21 (9), 1939–1946. Beck-Friis, B., Smars, S., Jo¨nsson, H., Kirchmann, H., 2001. Gaseous emissions of carbon dioxide, ammonia and nitrous oxide from organic household waste in a compost reactor under different temperature regimes. Journal of Agricultural Engineering Research 78 (4), 423–430. Chadwick, D.R., 2005. Emissions of ammonia, nitrous oxide and methane from cattle manure heaps: effect of compaction and covering. Atmospheric Environment 39 (1), 787–799. CIGR, 1984. Report of working group on climatization of animal houses. Commission Internationale du Ge´nie Rural. S.F.B.I.U., Aberdeen. Dulphy, J.P., Demarquilly, C., 1981. Proble`mes particuliers aux ensilages. In: Pre´vision de la valeur nutritive des aliments des ruminants, (Ed.) INRA publications, Versailles, pp. 81–104. Ekinci, K., Keener, H.M, Elwell, D.L., 2000. Composting short paper fiber with broiler litter and additives. Part I: effects of

ARTICLE IN PRESS 6842

J.-M. Paillat et al. / Atmospheric Environment 39 (2005) 6833–6842

initial pH and carbon/nitrogen ratio on ammonia emission. Compost Science and Utilization 8 (2), 160–172. Ekinci, K., Keener, H.M, Elwell, D.L., Michel, F.C., 2004. Effects of aeration strategies on the composting process: Part I. Experimental studies. Transactions of the ASAE 47 (5), 1697–1708. Eklind, Y., Kirchmann, H., 2000. Composting and storage of organic household waste with different litter amendments. I: carbon turnover. Bioresource Technology 74 (1), 115–124. Fangmeier, A., Hadwiger-Fangmeier, A., Van der Eerden, L., Ja¨ger, H.J., 1994. Effects of atmospheric ammonia on vegetation—A review. Environmental Pollution 86 (1), 43–82. Fukumoto, Y., Osada, T., Hanajima, D., Haga, K., 2003. Patterns and quantities of NH3, N2O and CH4 emissions during swine manure composting without forced aeration— effect of compost pile scale. Bioresource Technology 89 (1), 109–114. He, Y., Inamori, Y., Mizuochi, M., Kong, H., Iwami, N., Sun, T., 2000. Measurements of N2O and CH4 from the aerated composting of food waste. The Science of the Total Environment 254 (1), 65–74. Hellmann, B., Zelles, L., Palojarvi, A., QuingYun, B., 1997. Emission of climate-relevant trace gases and succession of microbial communities during open-windrow composting. Applied and Environmental Microbiology 63 (3), 1011–1018. Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., Van der Linden, P.J., Dai, X., Maskell, K., Johnson, C.A., 2001. Climate change 2001: the scientific basis. Report of the IPCC working group I. Ja¨ckel, U., Thummes, K., Ka¨mpfer, P., 2004. Thermophilic methane production and oxidation in compost. FEMS Microbiology Ecology in press. Kirchmann, H., 1985. Losses, plant uptake and utilisation of manure nitrogen during a production cycle. Acta Agriculturae Scandinavica. Supplementum No. 24. Kirchmann, H., Witter, E., 1989. Ammonia volatilization during aerobic and anaerobic manure decomposition. Plant and Soil 115 (1), 35–41. Kroeze, C., 1994. Nitrous oxide and global warming. The Science of The Total Environment 143 (2–3), 193–209. Liang, Y., Leonard, J.J., Feddes, J.J., McGill, W.B., 2004. A simulation model of ammonia volatilization in composting. Transactions of the ASAE 47 (5), 1667–1680. Mancinelli, R.L., 1992. Nitrogen cycle. Encyclopedia of Microbiology 3, 229–237.

Martins, O., Dewees, T., 1992. Loss of nitrogenous compounds during composting of animal wastes. Bioressource Technology 42 (1), 103–111. Morvan, T., Dach, J., Parnaudeau, V., 2001. Relationships between biochemical composition of manure and composts and their carbon and nitrogen transformation in soil. 11th Nitrogen Workshop. Reims, 153–154 (9–12 sep.). Mustin, M., 1987. Le compost, gestion de la matie`re organique. Edition Franc- ois Dubusc, Paris. Pare´, T., Dinel, H., Schnitzer, M., Dumontet, S., 1998. Transformations of carbon and nitrogen during composting of animal manure and shredded paper. Biology and Fertility of Soils 26 (1), 173–178. Peigne´, J., Girardin, P., 2004. Environmental impacts of farmscale composting practices. Water, Air, and Soil Pollution 153 (1), 45–68. Pel, R., Oldenhuis, R., Brand, W., Vos, A., Gottschal, J.C., Zwart, K.B., 1997. Stable-isotope analysis of a combined nitrification-denitrification sustained by thermophilic methanotrophs under low-oxygen conditions. Applied and Environmental Microbiology 63 (2), 474–481. Robin, D., 1997. Inte´reˆt de la caracte´risation biochimique pour l’e´valuation de la proportion de matie`re organique stable apre`s de´composition dans le sol et la classification des produits organomine´raux. Agronomie 17 (1), 157–171. Schlege, H.G., 1993. General microbiology, 7th ed. Cambridge University Press, New York. Shi, W., Norton, J.M., Miller, B.E., Pace, M.G., 1999. Effects of aeration and moisture during windrow composting on the nitrogen fertilizer values of dairy waste composts. Applied Soil Ecology 11 (1), 17–28. Sommer, S.G., 2001. Effect of composting on nutrient loss and nitrogen availability of cattle deep litter. European Journal of Agronomy 14 (1), 123–133. Sommer, S.G., Mfller, H.B., 2000. Emission of greenhouse gases during composting of deep letter from pig production—effect of straw content. Journal of Agricultural Science, Cambridge 134 (1), 327–335. Van Soest, P.J., 1963. Use of detergents in the analysis of fibrous feeds. II—A rapid method for the determination of fibre and lignin. Journal of Association of Analytical Chemistry 46 (1), 829–835. Wilshusen, J.H., Hettiaratchi, J.P.A., De Wisscher, A., SaintFort, R., 2004. Methane oxidation and formation of EPS in compost: effect of oxygen concentration. Environmental Pollution 129 (2), 305–314.